# GIT 1.5.0-rc2 compiler warning

3 messages from 2007-01-22 to 2007-01-22. Participants: SungHyun Nam, Junio C Hamano, Robin Rosenberg.
Thread: https://gitlist.dev/t/6465

## SungHyun Nam, 2007-01-22 01:08

Subject: GIT 1.5.0-rc2 compiler warning
Message-ID: <45B40E7E.3030706@gmail.com>
URL: https://gitlist.dev/e/45B40E7E.3030706%40gmail.com

```
Hello,

Use CYGWIN.

gcc -o utf8.o -c -O2 -DNO_D_TYPE_IN_DIRENT -DNO_SYMLINK_HEAD
-DNO_FAST_WORKING_DIRECTORY -DNO_IPV6 -DSHA1_HEADER='<openssl/sha.h>'
-DNO_STRCASESTR utf8.c
utf8.c: In function `wcwidth':
utf8.c:148: warning: comparison is always false due to limited range of
data type
utf8.c:148: warning: comparison is always true due to limited range of
data type
utf8.c:149: warning: comparison is always false due to limited range of
data type
utf8.c:149: warning: comparison is always true due to limited range of
data type
utf8.c: In function `reencode_string':
utf8.c:313: warning: passing arg 2 of `libiconv' from incompatible
pointer type


'gcc -E .... utf8.c' gives me:

# 325 "/bin/../lib/gcc/i686-pc-cygwin/3.4.4/include/stddef.h" 3 4
typedef short unsigned int wchar_t;


Regards,
namsh

```

## Junio C Hamano, 2007-01-22 01:13

Subject: Re: GIT 1.5.0-rc2 compiler warning
Message-ID: <7vy7nvyic4.fsf@assigned-by-dhcp.cox.net>
URL: https://gitlist.dev/e/7vy7nvyic4.fsf%40assigned-by-dhcp.cox.net
In-Reply-To: <45B40E7E.3030706@gmail.com>

```
I know.  Imported code.

I do not think we can do much on a platform where wchar_t is too
narrow and BMP only anyway.

```

## Robin Rosenberg, 2007-01-22 20:44

Subject: Re: GIT 1.5.0-rc2 compiler warning
Message-ID: <200701222144.34934.robin.rosenberg.lists@dewire.com>
URL: https://gitlist.dev/e/200701222144.34934.robin.rosenberg.lists%40dewire.com
In-Reply-To: <7vy7nvyic4.fsf@assigned-by-dhcp.cox.net>

```
måndag 22 januari 2007 02:13 skrev Junio C Hamano:
> I know.  Imported code.
>
> I do not think we can do much on a platform where wchar_t is too
> narrow and BMP only anyway.

What's wrong is the assumption that wchar_t corresponds to a complete unicode 
character. On windows, two wchar_t (code units) may me combined to encode a
unicode character outside of the BMP.

Without investigating further, I note that Java adds methods with an 'int' 
argument to the Character utility class to handle code points, vs the 
original 'char' that represents a code unit. (unlike C, an int is always 
32-bit in Java).

-- robin
 

```